Jour.

of Geophysics
January 2017, Vol. XXXVIII No.1
pp 57 to 61

Band Limited Impedance Inversion of Blackfoot Field, Alberta, Canada

S. P. Maurya and K. H. Singh
Department of Earth Sciences, Indian Institute of Technology, Bombay, Powai, Mumbai
Email: spmaurya@iitb.ac.in

Abstract
Poststack inversion transforms a single seismic data volume into acoustic impedance through integration of seismic
data and well data. The benefits of acoustic impedance obtained after inversion are well established for stratigraphic
interpretation in comparison to conventional seismic data analysis. This paper describes a seismic inversion method
which has been carried out in Hampson Russell software (version 9.0) named, BLIMP (Band Limited Impedance).
Band limited post stack seismic inversion are carried out with a 3D seismic data to derive the stratigraphic information.
The methodology of inversion includes statistical analysis of log derived acoustic impedance with P-wave velocity,
density and spectral analysis of seismic data, wavelet estimation, seismic to well tie, generation of initial 3-D geological
model. Use of band limited inversion has validated inverted impedance results. Present study is based on data from
Blackfoot field, Alberta, Canada. Band limited inversion results have helped in delineating the stratigraphic prospect in
the study area. This study demonstrates the effectiveness of using acoustic impedance volume for performing stratigraphic
interpretation and reservoir characterization. From this study, it is noticed that near vertical reflection two way time
1060 ms to 1065 ms have low acoustic impedance and low density, which is interpreted as potential hydrocarbon
saturated zone.

Keywards: Band limited inversion, Seismic Parameters, Stratigraphy of reservoir

Introduction contains both the low and the high frequency component,
are usually employed to constrain the deviation of the
Seismic inversion is a technique which extract subsurface solution from the initial-guess model. The final inversion
models of the physical characteristics of rocks and fluids. results therefore relies on the seismic data along with the
This techniques often uses seismic data with the integration additional information (well data) but are also dependent
of well log data to compute the physical properties of the on the inversion methods (Ferguson and Margrave, 1996).
subsurface (Pendrel et al.,2006). The physical parameters
of interest are the P-impedance, S-impedance, velocity, The study area is Blackfoot field located southeast of
density and porosity. Interpretation of seismic section alone Strathmore in Alberta, Canada. The 3C-3D seismic dataset
is very difficult, so attributes based on seismic inversion are was recorded in two patches in October 1996 by CREWES
utilized to improve the interpretation (Chen and Sidney, (Dufour et al., 1998). The first patch covers the clastic
1997). The acoustic impedance (AI) inversion methods of Glauconitic channel of the Lower Cretaceous which
post stack seismic data became popular when algorithms of represents sediment filled incised valley; and the second is
wavelet amplitude and phase spectra extraction became over the reef-prone Beaverhill lake carbonates (Simin et al.,
available (Lindseth, 1979). Seismic inversions generate 1996). Here the 3Dseismic data related to the Glauconitic
high-resolution images of the subsurface which enhanced channel and thirteen well logs are used for analysis.
the interpretation of seismic data and significantly reduced
the drilling risk and cost (Pendrel, 2006). Methodology
Bandlimited impedance inversion (BLIMP) is a type of
Seismic acoustic impedance inversion has several
post stack inversion techniques which utilised post stack
limitations, (1) the seismic frequency band is limited to about
seismic data to transform it into impedance, density and P-
12-80Hz, below 12Hz and above 80Hz frequencies are not
wave velocity. The bandlimited impedance method begins
available and therefore, the low- and high- frequency input
with specifying the relationship between the seismic trace
data for inversion methods are missing (Russell, 1999). (2)
and seismic impedance (Ferguson and Margrav, 1996). Thus,
The non-uniqueness of the seismic inversion results can lead
define the normal incidence reflection coefficient as:
to multiple possible geologic models which are consistent
Zj + 1 - Zj
with observations (Latimar et al., 2000). (3) The seismic rj = (1)
inversion method suffers from multiple issues like multiple Zj + 1 + Zj
reflections, transmission losses, geometric spreading and where Zj is seismic impedance of Jth layer, and rj is seismic
frequency dependent absorptions (Carrazzone et al., 1996). reflectivity of Jth and (J + 1)th interface.
For reducing these uncertainties, additional information
needed which comes from well log data. These well log data Solve avove equation for impedance of (j+1)th layer.
58 Maurya and Singh

2 rj 1 + rj Results and Discussion

Z j + 1 = Z j (1 + ) Zj ( ), (2)
1 - rj 1 - rj Band-limited impedance inversion technique is applied to
Impedance of nth layer if we know impedance of the Blackfoot post stack seismic data. Figure 1 shows initial
1st layer is: impedance model which was built by interpolating the
1 + r1 1 + r 2 1 + rn - 1 acoustic impedance from well locations into the all inlines
Z n = Z 1( )( )...( ) (3) and cross-lines. Two interpreted seismic horizons were used
1 - r1 1 - r 2 1 - rn - 1 as guide for the interpolation. To construct the impedance
The acoustic impedance for the first layer needs to be model A 12- Hz low-pass filter was applied to the model for
estimated from a continuous layer above the target area. In two reasons. First, the low frequency impedance trend is
this method, the impedance for the jth layer can thus be required in order to recover the low frequencies which is
calculated as follows: missing from the stacked seismic data. The second is the
impedances above 12-Hz frequencies should be only
j
1 + rk
Zj + 1 = Z 1 Õ ( ). (4) obtained from seismic data (Brown et al., 2004), and
k =1 1 - rk therefore this frequency band should be removed from the
Divide above equation (4) by impedance of 1st layer that is well-log data while building the starting impedance model
Z1 and take the logarithm on both side, (Carrazzone et al., 1996). One composit trace is extracted
from seismic data and inverted it into impedance, density
Zj + 1 j
1 + rk j
) + å In( ) » 2å rk (5) and P-wave velocity using band-limited impedance inversion
In(
Z1 k =1 1 - rk k =1
techniques. The inverted result at the well location was
compared to the original log value at well 01-17 as shown
The last step follows from an approximation for ln which is in Figure 2 respectivily. Correlation coefficient and RMS
valid only for small r. Now on solving equation (5) for Zj+1 Error were calculated between seismic and well log data.
we have: The correlation coefficient and RMS Error are shown in
Zj + 1 = Z 1exp(2å rk )
j
(6) figure 3 and figure 4 for all available wells.
k =1

Model the seismic trace as scaled reflectivity: Sk = 2r , k

.
then above equation becomes: γ
Zj + 1 = Z 1exp(γ å Sk )
j
(7)
k =1

The above equation thus integrates the seismic trace and

then exponentiates the result to provide an impedance trace.
Band-limited impedance method uses this equation to invert
the seismic trace in the same manner as Waters (1978), with
some modifications related to the pre-conditioning of the
required impedance estimate and the scaling of the seismic
trace. (Ferguson and Margrave, 1996)

Seismic Band-limited inversion technique is limited to

the same frequency range as the seismic data (Ferguson and
Margrave, 1996). Thatswhy, a low frequency initial model
is added to seismic inversion.

1) The initial model is derived by filtering the impedance

log from the wells. Interpolation is used to obtain a
3-D initial model using wells and seismic horizons.
2) Apply iterative equation to seismic traces to obtain
band- limited seismic inversion.
3) Final iterative inversion product is a combination of
the initial model with the bandlimited seismic AI.

A very important limitation of the band-limited

impedance inversion is that seismic data must be in zero
phase. The seismic pre-stack data could be transformed to (a)
zero phase by using extracted wavelet (Maurya and Singh, Fig. 1. P-Impedance model generated using high cut
2015b) filter of 12Hz frequency.
Band Limited Impedance Inversion 59

A cross section of the inverted impedance is shown in Figure The low density and P-wave velocity between 1060ms
5. The low impedances between 1060ms and 1065ms level and 1065ms are marked and highlightedby ellipse.
are clearly visible in this Figure which is highlighted by
ellipse. Inverted density and P-wave velocity were also
shown in figure 6 and figure 7 respectivily.

Fig. 5. Cross-section of inverted impedance. The low

(a) (b) (c) impedance zone is highlighted by the ellipse. Well log
Fig. 2. Comparision of (a) inverted impedance (Red impedance are shown along the column in the seismic
Curve), (b) density (Red Curve) and (c) P-wave section for well 01-17 and 09-17.
velocity (Red Curve) with original values from
well log data (Blue curve).

Fig. 3. Variation of correlation coefficient

with all boreholes.

Fig. 6. Cross-section of inverted density. The low density

zone is highlighted by the ellipse. Well log density are
Fig. 4. Variation of root mean square error shown along the column in the seismic section for well
with all boreholes. 01-17 and 09-17.
60 Maurya and Singh

Fig. 9. Crossplot of inverted and original

(well log) density.

Fig. 7. Cross-section of inverted P-wave velocity. The low

Pwavevelocity zone is highlighted by the ellipse. P-wave
Fig. 10. Crossplot of inverted and original (well log)
velocity from well log are shown along the column in the
P-wave velocity.
seismic section for well 01-17 and 09-17.
Conclusion
Figure 8 shows the cross-plot of original impedance and
inverted impedance. The scatter data lies near to best fitt The present study used the Blackfoot 3D seismic data
line which shows the inverted results is agrees with the for analysis of Band-limited impedance inversion techniques.
original results. Figure 9 shows cross-plot of original density Post stack PP seismic gathers are utilized to perform seismic
and inverted density whereas figure 10 shows cross-plot of inversion methods for estimation of impedance, density and
original P-wave velocity and inverted P-wave velocity. The P-wave velocity volume. These inverted volume used for
scatter data lies near to best fit line in both the crossplots characterisation of sand channel. The final inverted sections
which shows the inverted results is agrees with the original showed good images within the time-depth ranges of 800ms
results. to 1300ms, where the target Glauconitic channel is located.

The inverted result shows variation of impedance, density

and P-wave velocity along with the seismic section. A low
impedance zone is estimated between 1060ms and 1065ms.
At the same time interval low density and low P- wave
velocity is found which confirm the presence of sand channel
in this zone. The average synthethetic correlation for all
thirteen wells availabe in the study area is 0.887 which shows
good correlation between attributes calculated from seismic
data and well log data. The average RMS Error for all thirteen
wells is 963.842(m/s)*(g/cc). Crossplot between inverted
attributes estimated from seismic data and attributes
estimated directly from well log data shows most of data
Fig. 8. Crossplot of inverted and original (well log) points lies near to best fit line which shows inverted results
Pimpedance. are good and comparable with the well log data.
Band Limited Impedance Inversion 61

Acknowledgments Lindseth, Roy O. ”Synthetic sonic logs-a process for

stratigraphic interpretation.”Geophysics 44.11979:
I express my gratitude to my colleagues for the support 3-26.
abd fruitfull discussions. This work was facilitated by
software and data grants from Hampson-russel Limited Maurya, S. P. and Singh, K. H., LP and ML sparse spike
(CGG Veritas), without their help this work would not be inversion for reservoir characterization-a case study
possible. from blackfoot area, alberta, canada. In 77th EAGE
Conference and Exhibition 2015, 2015a.
References
Maurya, S. P. and Singh, K. H., Estimation of seismic
Brown, Alistair R., et al. Interpretation of three-dimensional p ar a met e r s f r o m p r e -s t ac k in v er s i o n . I n 2 n d
seismic data. Tulsa: American Association of Petroleum international conference on computational and
Geologists, 2004. experimental sciences and engineering, Antalya,
Turkey, 2015b.
Carrazzone, James J., et al. ”Method for deriving reservoir
lithology and fluid content from pre-stack inversion of Pendrel, John. ”Seismic inversiona critical tool in reservoir
seismic data.” U.S. Patent No. 5,583,825. 10 Dec. 1996. characterization.” Scandinavian oil-gas magazine
Chen, Quincy, and Steve Sidney. ”Seismic attribute 5.6 2006: 19-22.
technology for reservoir forecasting and monitoring.”
The Leading Edge 16.5 (1997): 445-448. Robinson, Enders A. Seismic Inversion & Deconvolution:
Classical methods. Vol. 4. Geophysical press, 1984.
Dufour, J ocelyn, et al. ”Integrated geolo gical and
geophysical interpretation of the Blackfoot area, Russell, Brian. ”Comparison of poststack seismic inversion
Southern Alberta.” 1998: 234-236. methods.” SEG Technical Program Expanded Abst. No.
10. 1999.
Ferguson, Robert J., and Gary F. Margrave. ”A simple
algorithm for band-limited impedance inversion.” Russell, Brian H. Introduction to seismic inversion methods.
CREWES annual 1996. Vol. 2. Society of Exploration Geophysicists, 1988.

Latimer, Rebecca Buxton, Rick Davidson, and Paul Van Torres-Verdin, Carlos, et al. ”Trace-based and geostatistical
Riel. ”An interpreter’s guide to understanding and inversion of 3-D seismic data for thin-sand delineation:
working with seismic-derived acoustic impedance An application in San Jorge Basin, Argentina.” The
data.” The leading edge 19.3 2000: 242-256. Leading Edge 18.9 1999: 1070-1077.